Dependency updates

[christophe-david]

This PR cleans up and updates user dependencies of FAST-OAD:

• jupyterlab 3.0.18 is now a required dependency (not an extra). ipysheet 0.5 or above and plotly 5.x or above are now required to ensure they are usable in Jupyterlab without any need for extra build.
  • Actually, plotly 5.2.1 or above is required because it brings some API changes that are used for the replacement of matplotlib (see below)
  • Process for building Binder image has been updated, since it does not require a postbuild step anymore.
• matplotlib has been moved from user dependencies to dev dependencies. It is used only in tests (and these parts are just visual checks that are currently commented out), so it looks irrelevant to make users install it. It could be replaced by plotly, but plotly needs kaleido to write image files, and kaleido is a huge library that does not look very mature.
• jsonschema 4.x is now allowed.
• wop minimum required version has been updated to 1.16.1.1 (needed to be compatible with OpenMDAO 3.10)
• scipy minimum required version has been updated to 1.4.1 (lower versions make some integration tests fail. I guess optimizers do not work exactly the same)
• Poetry lock file has been locked file to a set of dependencies that aims at being minimal while being compatible with Python 3.9
  • Since the "watchman tests" check what happens with latest authorized versions of all libraries, it appeared to me that it is better to have our development environment based on "oldest" possible versions.
  • To be noted: poetry updatewill not lock on latest versions of pandas, numpy and scipy because of the expressed Python requirement in pyproject.toml ("^3.7.0", equivalent to ">=3.7.0, <4.0"):
  • pandas won't update above 1.1.5 (latest is 1.3.3) because it requires at least Python 3.7.1
  • numpy stalls at version 1.21.1 and do not get to 1.21.2 because this version forbids Python >=3.11
  • scipy stalls at version 1.6.1 (latest is 1.7.1) because versions above 1.6.1 forbid Python >=3.10
  • of course, these limitations for numpy and scipy will eventually be removed with future versions
  • For the record, our "watchman tests" do a simple pip install rather than a poetry update, and that's good, because this way, latest versions of poetry, numpy and scipy are installed.

[codecov[bot]]

Codecov Report

Merging #392 (673515d) into master (d8bc9f1) will not change coverage. The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #392   +/-   ##
=======================================
  Coverage   90.37%   90.37%           
=======================================
  Files         200      200           
  Lines        8174     8174           
  Branches      850      850           
=======================================
  Hits         7387     7387           
  Misses        636      636           
  Partials      151      151           

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update d8bc9f1...673515d. Read the comment docs.